Inference
EN
EuroEval

EuroEval Dutch

How well a model handles Dutch, as the unweighted mean of EuroEval's primary metric over its 9 Dutch datasets (dbrd, conll_nl, dutch_cola, squad_nl, wiki_lingua_nl, include_nl, multiloko_nl, winogrande_nl, duidelijke_taal). The primary metric differs by task — MCC for classification, knowledge and common-sense reasoning, micro-F1 without MISC for entity recognition, F1 for reading comprehension, chrF++ for summarization, METEOR for simplification — and is always the first of the two the board prints. Only models measured on every dataset are included, so each mean covers the same ground. Most rows are the validation split; some older open-weight rows are the test split, which EuroEval ranks in the same table.

ModelsModels with a result.168
Top resultBest result on this benchmark.74.7%GPT-6 Astra
Top-3 spreadPoints from first to third.1.6 pts
YearYear of release.—

Result and price

38 of 175 results

Results

175 results
#ModelResultThe score from the source.
01GPT-6 Astra74.7% ±0.3
02GPT-6.1 Sol74.0% ±0.4
03Gemini 3.7 Flash73.1% ±0.4
04Gemini 3.6 Flash71.8% ±0.3
05Gemini 3 Flash Preview Thinkingthinking effort69.4% ±0.4
06GPT-5.6 Sol68.7% ±0.4
07GPT-6 Sol67.5% ±0.4
08GLM-5.3-Flash66.0% ±0.4
09DeepSeek Flash No Thinkingno-thinking effort65.4% ±0.5
10Claude Sonnet 4.5 Thinkingthinking effort65.3% ±0.4
11GPT-5.6 Terra65.0% ±0.5
12DeepSeek Flash Lowlow effort64.3% ±0.4
13Qwen3.6 27B FP864.0% ±0.4
14Grok 4.1 Fast (Reasoning)63.8% ±0.4
15Qwen3.5 35B A3B FP863.3% ±0.6
16GPT-5.4 minihigh effort63.1% ±0.4
17GPT-5.4 minimedium effort63.0% ±0.5
18Gemma 4 31Bgemma-4-31B-it62.7% ±0.4
19GPT-4.1Above fitFar above this model's index.62.6% ±0.5
20Claude Sonnet 4.5 No Thinkingno-thinking effort62.5% ±0.4
21Qwen3.8-27B62.1% ±0.5
21GPT-5.4 minilow effort62.1% ±0.4
23Gemini 3.1 Flash Lite Preview61.4% ±0.4
24GPT-5.4 Nonenone effort60.8% ±0.4
25GPT-5 mini60.1% ±0.4
26Claude Sonnet 4.659.7% ±0.4
27Gemini 3.5 Flash-Lite59.6% ±0.5
28GPT-5.6 Luna59.3% ±0.6
29GPT-6 Luna59.0% ±0.4
30GPT-OSS 120B57.8% ±0.4
31Mistral Small 3.1 24B57.6% ±0.3
32Mistral Small 3.2 24B57.4% ±0.3
32Gemini 2.5 Flash Lite Thinkingthinking effort57.4% ±0.5
34Gemma 4 12B IT57.3% ±0.5
35Mistral Small 3.1 24B Base57.2% ±0.5
36Cogito V1 Preview Qwen 32B56.7% ±0.2
37Gemma 3 27B Pt56.5% ±0.4
38Gemma 4 26B A4Bgemma-4-26B-A4B-it56.4% ±0.5
39Qwen3 32B No Thinkingno-thinking effort56.3% ±0.3
40GPT-5.256.2% ±0.5
41Qwen3.5 9B Base55.6% ±0.7
42Ministral 3 14B Base55.5% ±0.6
42NVIDIA Nemotron 3 Nano 30B A3B BF1655.5% ±0.6
44GPT-5.4 Mini Nonenone effort55.3% ±0.5
45Qwen3 14B No Thinkingno-thinking effort54.4% ±0.3
46NVIDIA Nemotron 3 Nano 30B A3B FP854.2% ±0.5
47Qwen3 4B Thinking53.8% ±0.4
47Bielik 11B V3.0 Instruct53.8%
49GPT-5 nano53.7% ±0.5
50Cogito V1 Preview Qwen 14B53.5% ±0.2
50Nemotron Cascade 2 30B A3B53.5% ±0.6
52Claude Haiku 4.553.4% ±0.5
52Gemma 3 12B Pt53.4% ±0.4
54GLM-4.7-Flash53.3% ±0.6
55GPT-5.4 nanomedium effort53.1% ±0.4
56GPT-5.4 nanohigh effort53.0% ±0.5
57Ministral 3 14B Instruct52.6% ±0.5
58Gemma 3 12B52.4% ±0.3
58Gemini 2.5 Flash Lite No Thinkingno-thinking effort52.4% ±0.5
60Qwen3.5 9B52.2% ±0.5
60Apertus V1.5 70B52.2% ±0.6
62Gemma 4 31Bgemma-4-31B51.8% ±0.7
63GPT-5.4 nanolow effort51.6% ±0.4
64EuroLLM 22B Instruct 251251.5% ±0.3
65Bielik Minitron 7B V3.0 Instruct51.4%
66Ministral 3 14B (Reasoning)50.7% ±0.6
66Ministral 3 8B (Reasoning)50.7% ±0.3
68Ministral 3 8B Base50.1% ±0.7
69NVIDIA Nemotron 3 Nano 30B A3B Base BF1650.0% ±0.7
70GPT-OSS 20Blow effort49.9% ±0.3
71Gemma 4 E4B IT49.8% ±0.5
72Qwen3 8B No Thinkingno-thinking effort49.7% ±0.3
73Ministral 3 8B Instruct48.6% ±0.6
74Qwen3.5 4B Base48.3% ±0.9
75EuroLLM 9B Instruct 251247.1% ±0.6
76Llama 3.1 8B Instruct46.8% ±0.3
77EuroLLM 22B 251246.7% ±0.6
78Llama 3.1 Tulu 3 8B SFT46.5% ±0.4
79Qwen3 4B No Thinkingno-thinking effort46.2% ±0.3
80Nesso 4B45.8% ±0.3
81Cogito V1 Preview Llama 8B45.7% ±0.5
82Aya Expanse 8B45.5% ±0.7
83Gemma 4 26B A4Bgemma-4-26B-A4B45.3% ±0.7
84Gemma 4 12B44.9% ±0.7
85Qwen3.5-4B44.6% ±0.7
85Luciole 23B Instruct 1.144.6% ±0.6
87EuroLLM 9B Instruct44.5% ±0.4
88Domyn Small V1.044.3% ±0.6
89Llama 3.1 8B43.8% ±0.4
90AMALIA 9B DPO43.3% ±0.6
91Gemma 4 E2B IT43.1% ±0.5
92AMALIA 9B SFT43.0% ±0.4
92EuroLLM 9B43.0% ±0.5
94Gemma 3 4B Pt41.6% ±0.4
95Ministral 3 3B (Reasoning)41.5% ±0.4
96Gemini 3 Flash Preview No Thinkingno-thinking effort41.4% ±0.5
97Exaone 4.0 32B41.2% ±0.6
98GPT-OSS 20Bhigh effort40.5% ±0.6
98GPT-5.4 Nano Nonenone effort40.5% ±0.5
98Ministral 3 3B Instruct40.5% ±0.4
101Qwen3.5 2B40.2% ±0.6
102Luciole 23B Base40.0% ±0.6
103Gemma 3 4B39.8% ±0.6
104Apertus V1.1 4B38.4% ±0.6
105Llama 3.2 3B Instruct38.2% ±0.4
106Luciole 8B Instruct 1.137.5% ±0.7
107SmolLM3 3B Base37.4% ±0.4
108Mellum2-12B-A2.5B-Instruct37.1% ±0.6
109Qwen3 1.7B No Thinkingno-thinking effort36.5% ±0.3
109DFM Mimir36.5% ±0.5
111LFM2.5 8B A1B Base36.3% ±0.7
112DFM Mimir V1.535.9% ±0.5
113Apertus V1.1 4B Instruct35.7% ±0.5
113Qwen3.5 2B Base35.7% ±0.7
115Occiglot 7B Eu535.2% ±0.4
116Mellum2 12B A2.5B Instruct SFT35.1% ±0.6
116Oellm 9B 128K Theta32m V335.1% ±0.7
118Occiglot 7B Eu5 Instruct34.8% ±0.4
119Luciole 8B Base33.6% ±0.7
120Mellum2 12B A2.5B Thinking SFT33.4% ±0.9
121Gemma 4 E4B33.2% ±0.8
122Aya 23 8B33.1% ±0.7
123NVIDIA Nemotron 3 Nano 4B FP832.7% ±0.8
124Cogito V1 Preview Llama 3B32.3% ±0.3
125Qwen3.5 0.8B31.9% ±0.6
126SmolLM3 3B No Thinkingno-thinking effort31.6% ±0.3
127NVIDIA Nemotron 3 Nano 4B BF1631.5% ±0.4
128Mellum2-12B-A2.5B-Thinking31.2% ±0.9
129Trustllm V2 8B Midtrain30.5% ±0.4
130Teuken 7B Instruct V0.630.1% ±0.5
131Lizzy 7B28.8% ±0.6
132Trustllm V2 8B28.6% ±0.5
133Salamandra 7B Instruct28.1% ±0.3
134Mellum2 12B A2.5B Base Pretrain27.6% ±1.0
135Apertus V1.1 1.5B27.1% ±0.7
136Qwen3.5 0.8B Base27.0% ±0.7
137Bielik 1.5B V3.0 Instruct26.6% ±1.4
138Salamandra 7B26.2% ±0.4
139Llama 3.2 1B25.3% ±0.5
140Qwen3 0.6B24.8% ±0.4
140Trustllm V2 8B SFT 8K24.8% ±0.2
142Qwen3 0.6B No Thinkingno-thinking effort24.6% ±0.4
143Gemma 3 1B Pt24.4% ±0.7
144YuLan Mini Instruct24.3% ±0.5
145Luciole 1B Instruct 1.124.2% ±0.7
146Gemma 4 E2B24.1% ±0.7
147Trustllm V2 8B SFT Mixb2 Exp24.0% ±0.3
148Apertus V1.1 0.5B23.7% ±0.8
149Trustllm V2 8B SFT23.4% ±0.2
150Luciole 1B Base23.3% ±0.9
151Apertus V1.1 0.5B Instruct22.5% ±0.7
152Apertus V1.1 1.5B Instruct22.2% ±0.8
153Gemma 3 1B IT21.5% ±0.3
154LFM2.5-8B-A1B21.1% ±0.7
155Pleias 1.2B Preview20.8% ±0.6
156LFM2 1.2B18.2% ±0.4
157EuroMoE 2.6B A0.6B Instruct Preview17.2% ±0.3
158Llama 3.2 1B Instruct15.0% ±0.5
159SmolLM2 360M14.2% ±0.5
160Pleias 350M Preview14.0% ±0.5
161Gemma 3 270M13.6% ±0.6
162LFM2 700M13.3% ±0.4
163Falcon H1 Tiny Multilingual 100M Instruct12.8% ±0.3
164Falcon H1 Tiny Multilingual 100M Base12.3% ±0.4
165LFM2 350M10.4% ±0.5
166Ling Mini 2.09.9% ±0.6
166SmolLM2 135M9.9% ±0.4
166Gemma 3 270M IT9.9% ±0.5
169Baguettotron No Thinkingno-thinking effort7.8% ±0.5
170SmolLM2 360M Instruct7.7% ±0.4
171SmolLM2 135M Instruct7.5% ±0.3
172Marco Mini Global Base3.5% ±0.6
173Monad2.1% ±0.3
174Muse 3B0.4% ±0.5
175Teuken 7B Base V0.60.0%